VideoUNetArchitectureProfile

Enums · AiDotNet.Enums

Topology profiles supported by the shared video U-Net predictor.

Fields

Field Summary
VideoUNetArchitectureProfile Generic General video-diffusion U-Net behavior retained for existing callers.
VideoUNetArchitectureProfile UpscaleAVideo Released Upscale-A-Video topology: two spatial ResNets per down block, three per up block, all-stage convolutional temporal modules, and the zero-initialized temporal attention embedded in Transformer3D blocks.
int value__